A COMPARISON AMONG DIFFERENT OPTIMIZATION LEVELS IN 3D MULTI-SENSOR MODELS.
A TEST CASE IN EMERGENCY CONTEXT: 2016 ITALIAN EARTHQUAKE
F. Chiabrando
a
, G. Sammartano
a
, A. Spanò
a a
Politecnico di Torino, DAD, 10129, Torino, Italy - filiberto.chiabrando, giulia.sammartano, antonia.spanopolito.it
Commission II KEY WORDS:
emergency mapping, 3D documentation, Cultural Heritage risk, multi-scale modelling, multi-sensor approach,
UAV, SfM photogrammetry, SLAM, information extraction, data filtering, point cloud post-processing
ABSTRACT: In sudden emergency contexts that affect urban centres and built heritage, the latest Geomatics technique solutions must enable the
demands of damage documentation, risk assessment, management and data sharing as efficiently as possible, in relation to the danger condition, to the accessibility constraints of areas and to the tight deadlines needs. In recent times, Unmanned Vehicle System
UAV equipped with cameras are more and more involved in aerial survey and reconnaissance missions, and they are behaving in a very cost-effective way in the direction of 3D documentation and preliminary damage assessment. More and more UAV equipment
with low-cost sensors must become, in the future, suitable in every situation of documentation, but above all in damages and uncertainty frameworks. Rapidity in acquisition times and low-cost sensors are challenging marks, and they could be taken into
consideration maybe with time spending processing. The paper will analyze and try to classify the information content in 3D aerial and terrestrial models and the importance of metric and non-metric withdrawable information that should be suitable for further uses,
as the structural analysis one. The test area is an experience of Team Direct from Politecnico di Torino in centre Italy, where a strong earthquake occurred in August 2016. This study is carried out on a stand-alone damaged building in Pescara del Tronto AP, with a
multi-sensor 3D survey. The aim is to evaluate the contribution of terrestrial and aerial quick documentation by a SLAM based LiDAR and a camera equipped multirotor UAV, for a first reconnaissance inspection and modelling in terms of level of details,
metric and non-metric information.
Corresponding author
1. INTRODUCTION
In recent years, we can affirm that the increasing use of remoted piloted system RPAS equipped with cameras have already
improve its role in competitiveness and efficiency in surveying operation on the field. The phenomenon is covering nowadays
many
application domains
geomatics, geotechnics,
archaeology, forestry, structural analysis, etc. The current aspect is connected to the diffusion of a large number of
operators not expert people as well and concurrently to the efficient
image-matching algorithm
based of
SfM photogrammetry even more suitable for 3D information
extraction from images and frames videos acquired by compact cameras embedded in low cost commercial drones. This
technique is moving in a very profitable way in many contexts, and mostly popular in the direction of 3D documentation and
monitoring Cultural Heritage sites Chiabrando et al, 2016a; Ruiz Sabina et al. 2015; Remondino et al., 2011; Lerma et al.,
20122. The use of drone is equally diffuse in these last years in urban areas involved in natural disaster for preliminary search
and rescue, building damage assessment BDA yet Fernandez Galarreta, J. et al, 2015; Meier, P. 2016. This is sometimes
preferred than the more traditional vertical images from remote sensing data. Due to their scale, their geometric configuration
and ultimately their intrinsic features, satellite imaging does not satisfy the requirements of details and information. Lemonie et
al, 2013; Rastiveis et al., 2013
; Gerke, Kerle, 2011. The first data acquisition phase, to be done as quickly as
possible after a disaster, is ordinarily carried out in person with manty efforts by technicians in damaged sites and is a heavy
time-consuming operation. Rapidity in acquisition times must be effectively balanced with post processing times, as well as
the time-cost ratio must be successful, in favor of low-cost sensors with their top efficiency, as the best possible
compromise between timeliness and accuracy Lemonie et al, 2013. It must maximize, ultimately, the density of data and the
metric data extraction from 3D models processed ex post, without neglecting productivity in ex ante data acquisition in
emergency circumstances, where the practicability of spaces connected to the high risk they could cause, can adversely affect
the quality of data. Moreover, 3D models already are, and could be increasingly in the near future, an essential platform of dense
information for interdisciplinary teamwork on the object of study, which will be analyzed in remote, in a second step, for
many purposes. They could be, for example: emergency measures of rescue and second aid for civil protection and
firefighters, or for damage detection and structural assessment, and for planning a structural strengthening project, or even
historical documentation studies as well as restoration analysis and intervention.
This paper wants to base its proposal on the possibility of interpretation and classification of metric and non-metric
information starting from high-scale photogrammetric models and data extraction from them. This process is planned in the
direction of quick surveys and targeted operation helpful to assess the state of conservation and seismic damage in these
types of building belonging to a precarious context. The selection of survey methods depends primarily on these issues,
and then it is influenced by needs of information detection on those models. Metric data and its extraction constitutes the
This contribution has been peer-reviewed. doi:10.5194isprs-archives-XLII-2-W3-155-2017
155
unavoidable base on which to base non-metric information, as visual and qualitative ones. Here we will focus the analysis on image-
based techniques for quick 3D modelling, using multi point of view image acquisition by RPAS and exploiting them together with
integration of terrestrial acquisition, for Cultural Heritage documentation, specifically in disaster areas, as the one of Pescara
del Tronto, in center Italy. Here a strong earthquake occurred in August 2016, and where current ongoing seismic shocks still take
place. The moment magnitude of this event is listed as 6.0 by INGV Istituto
Nazionale di
Geofisica e
Vulcanologia, http:cnt.rm.ingv.it
, that places the hypocentral depth of the event at 8 km. The 24 August event was followed by two additional
earthquakes: a moment magnitude 5.9 event on 26 October, and a moment magnitude 6.5 event the largest event on 30 October. Here
it requires urgent quick survey, damage assessment, and measures of safety and consolidation.
2. MULTI-SENSOR MODEL COMPLEXITY